IETF 85 Plenary Minutes Atlanta, GA, USA Wednesday, 7 November 2012 Minutes by Cindy Morgan, IETF Secretariat 1. Welcome Russ Housley welcomed the audience to the IETF 85 plenary. Based on community feedback after previous plenaries, the Operations and Administrative plenary was combined with the Technical Plenary, and a format with reduced reporting was used. The reports include links for interested parties to find additional information if desired. 2. Host Presentation http://www.ietf.org/proceedings/85/slides/slides-85-iesg-opsandtech-12.pptx Jean-François Mulé accepted the host plaque on behalf of the North American Cable Industry. 3. Reporting 3.1. IETF Chair http://www.ietf.org/proceedings/85/slides/slides-85-iesg-opsandtech-13.ppt 3.2. IAOC Chair and IAD Report http://www.ietf.org/proceedings/85/slides/slides-85-iesg-opsandtech-14.pptx 3.3. IETF Trust Chair Report http://www.ietf.org/proceedings/85/slides/slides-85-iesg-opsandtech-15.ppt 3.4. IAB Chair Report http://www.ietf.org/proceedings/85/slides/slides-85-iesg-opsandtech-3.pdf 3.5. IRTF Chair http://www.ietf.org/proceedings/85/slides/slides-85-iesg-opsandtech-0.pdf 3.4. NomCom Chair http://www.ietf.org/proceedings/85/slides/slides-85-iesg-opsandtech-16.ppt 4. Awards and Recognition http://www.ietf.org/proceedings/85/slides/slides-85-iesg-opsandtech-17.ppt Jun Murai introduced the Itojun Service Award and announced this year's winners: John Jason Brzozowski, Donn Lee and Paul Saab. The winners thanked the committee and encouraged the community to keep deploying IPv6. 5. Technical Topic: Internet Performance Measurement Alissa Cooper introduced the technical topic for the plenary, Internet Performance Measurement. Two presentations were delivered: - Sam Crawford, "SamKnows" http://www.ietf.org/proceedings/85/slides/slides-85-iesg-opsandtech-7.pdf - Henning Schulzrinne, "Large-Scale Internet Measurements for Data- Driven Public Policy" http://www.ietf.org/proceedings/85/slides/slides-85-iesg-opsandtech-8.pdf At the end of the presentations, the microphones were opened for questions from the audience. A summary of the Q-and-A session follows: JUN MURAI: I have question for Henning. I agree with what you said about measurement being key, and it's going to be important. We've been working for a long time in the past on the measurement of the Internet, and the measurement work in research has been confused in many ways. There is the business secrets concern, and also the government concern. What is your suggestion, being the FCC in America, to have non-Americans join you? What is your encouragement to the rest of the world? HENNING SCHULZRINNE: It depends on which audience you're talking to. We'd be happy to talk to any of our colleagues, so if you want to send them our way we can certainly talk about our experience. I believe we have demonstrated two things, namely that it doesn't have to be an adversarial relationship; it's not a regulator versus the ISP type of thing. It can be a collaborative relationship where everybody benefits. And, it is something where we can show that it actually helps performance improvement, which presumably everybody benefits from. If we do our job here, even for smaller countries where it might be difficult to set up a completely new infrastructure, hopefully they can all just ride on larger efforts. OLAF KOLKMAN: I was very happy that your presentation drove home the point that the goal is that good policy needs good data. That's an important message. But what is there that we can learn by doing good measurement with respect to Internet architecture? Is there anything you've seen in your measurements that we can use in our protocol and architecture development in the IETF? SAM CRAWFORD: Our focus so far has been on over-the-top measurements, looking at the various aspects of performance, rather than features. But it's something we're exploring right now. There are a couple of initiatives we are looking at. One is around, not just DNSSEC support, but DNSSEC compliance or grading. And another is around IPv6. We've already been doing some extensive IPv6 measurements in Singapore. We've seen some interesting differences between v4 and v6. Not just performance, but how the different networks behave with each other. HENNING SCHULZRINNE: I'm not sure we've learned anything inherently about Internet architecture as an abstract entity, but what we do learn is about what really happens to the stuff that we build around here, in the real world. The assumptions that some of us may have are just no longer reflective of reality, and so it provides us also an indication of the kind of spread of parameters that you see in the real world. That might lead us to have better diagnostics tools in some of the protocols that we build, it might make us think about making it easier for protocols to be measurable, and it might (as we progress documents) give us an indication that certain features are particularly problematic and we might want to look at them again. ERIC VYNCKE: Most of the measurement right now is in the network layer. You talk about IPv6, but you don't talk about the doom days of CGN. What is the support of non-TCP and UDP in the transport layer? Do you measure only the subscribers behind CGN? And if they are behind CGN, how many TCP and UDP sessions do you connect? SAM CRAWFORD: No, we're not measuring how many subscribers there are in those areas. With all of these studies from regulators, one of the first things we do is a sample plan highlighting which areas we're looking for from customers, and this goes to where the ISPs have their customers. We build the panel around that, so if there is a concentration of users in one area, then that is one of the aspects we're measuring. With regards to how many TCP and UPD sessions, it varies by region. We do typically use multiple concurrent sessions in order to overcome limitations. ALISSA COOPER: But you're not measuring the maximum achievable. HENNING SCHULZRINNE: Let me see if I can twist the question a little bit. We do TCP and UDP, but there are other protocols out there that people want to use, SCTP being one, for example. I think this an emerging area where there is just starting to be an awareness that broadband is more than speed. And from a regulatory perspective, because it's easier to explain speed to policy makers who might be lawyers and economists that like nice round numbers, as long as it's broadband and has some speed, the Internet must all be the same. Since we're no longer concerned with just the availability of broadband, but what do you do with it, innovation, all of that. Clearly, if you look at our Internet principle work, we call out the need for networks that are not constrained in their abilities. Increasingly, this is not something that we have a work plan for. We will be looking at how to characterize the evolution of what you can do with the network as applications emerge. Much of what we do is driven by the questions we get asked. The measurement collaborative is an open one. If interested parties have something to contribute, we encourage them to participate. We encourage people to approach us with the issues they see in ways that may not be consumer-obvious. MICHAEL RICHARDSON: Yes, performance is not the only thing. Are you measuring latency? Are you able to measure it with your current system? From what I can tell, that power-boost stuff is actually a cause of trouble rather than a feature. It's great to hear that performance is more than speed, but what numbers can we really use at the end, and can you measure them at this point? SAM CRAWFORD: Yes, we are measuring latency. We're measuring effectively to the nearest point of Internet exchange, to the nearest hand-off from the ISP. So yes, we're capturing that, and a whole bunch of other metrics as well. HENNING SCHULZRINNE: Are you talking about buffer-bloat issues? MICHAEL RICHARDSON: Yes, exactly. HENNING SCHULZRINNE: So yes, we have one measurement that is the latency under load, which is an initial attempt to identify the presence if not necessarily characterize the full extent of this type of load-sensitive latency, as opposed to the standard transmission layer, DSL and coax type of latencies that you measure. MICHAEL RICHARDSON: So you're presenting two speeds, is what you're saying? Basic speed and speed under load? HENNING SCHULZRINNE: Yes. And that's probably an area where additional work here [in the IETF] would be most helpful. ROBBIE SIMPSON: I think the area of Internet measurement research and standards work has been going on for quite some time. Some time ago I started a project called Net-E at Home, and I thought we were sitting on a gold mine of data at the time, but unfortunately we couldn't share the data because of the privacy issues. Has the FCC or any other regulatory body published any guidelines on what measurements you think are safe to share? HENNING SCHULZRINNE: We've thought a lot about privacy issues and struggled internally about what you can reasonably disclose, because we want to disclose pretty much whatever we measure, including IP addresses and everything. So we did have some lawyers look at that. We arrived at a conclusion which I'm not sure I can generalize, simply because national laws matter. In the U.S., national laws are often not well- defined for these types of things. There just isn't a law to address that. What is a potential risk and impact? How much informed consent do you have? So we intentionally pick active measurements as opposed to passive measurements. We only measure when there's no other activity going on. We pick what you might call the easy part. We are currently working with a number of privacy experts for the mobile piece because that is an order of magnitude harder. You have the conflict between needing the location and not wanting to disclose the location as part of your measurement data. BERT WIJNEN: I think this is a good start. It's unfortunate that Henning's presentation was very U.S.-centric, but at least we get the idea of how you guys do it over here, and we can see how that applies at our place. I think people have already mentioned that we need to do more than just bandwidth management. In the BOF that we're going to have, we can probably find out if we can define some standard protocols or data formats. So, a shameless plug for the RIPE NCC Atlas Project, which already does traceroutes, pings, DNS queries, HTTP GETs and what- have-you. For example, we did see data with Hurricane Sandy. You can also do user-defined measurement, but of course that gets into security and all that stuff. They user-defined measurements that do allow people to see something happening in the small scale in their environment that allow them to set up a quick measurement that will check things around the area where they see things happening. HENNING SCHULZRINNE: I look forward to becoming aware of other projects. There's a number of these efforts. Every country has at least one project. The EU seems to like to do that. I'm hoping that if we start that LMAP effort here that the experience can be introduced. I apologize for being U.S.-centric, but I don't want to pretend to speak for other policy bodies. BERT WIJNEN: Right, but if we can define interfaces to all the data, then that will allow us to correlate measurements that SamKnows does with RIPE Atlas stuff, etc. And that would give us a much better insight into the performance for the user, but also if there are any problems, wherever they may be. ARTURO SERVIN: Last week, there was a presentation on some of the research we are doing in Latin America, and Patrik Fältström went to the mic and said that this was the 5th presentation he'd seen on measurement in the last week. Initially I thought he was exaggerating, but it's been a week, and this was my 4th one. Indeed, it seems like there are a lot of people doing measurement, but there is no data, so we have to do our own research. So this is more a comment, that we need to start finding some common ground amongst each other, because there's a lot of money and effort expended here. HENNING SCHULZRINNE: Certainly we hope to make it easy for people to do their own measurements by taking care of the boring bits, so that researchers can focus on the interesting statistical issues. That's certainly one of my goals. PATRIK WALLSTROM: At .SE, we have been doing these kind of measurements for the past 1.5 years. We have 30 or so probes that we have on different connections, and we've been measuring from all these interconnecting connections. What we have found is that we see a lot of filtering in bittorrent and VoIP protocols, but for different reasons than you might think. One example is that you see the filtering only where the traffic is expensive, for example traffic going though a trusty operator. One other example is on your mobile network--on some of the best mobile networks in Sweden, you see the most filtering. So that's data that you have to analyze and see the reasons for. HANNES TSCHOFENIG: I would like to touch on the architectural aspects. As a protocol designer, you have to take limitations in the deployment into account. There are different academic papers on the inability to get, for example, the IP options through the network, ECN, etc. Some believe you have to use HTTPS nowadays to get any application protocol to work in all the different networks. Have you done some measurements that would actually give us any large-scale data on this? Because I couldn't find anything. SAM CRAWFORD: This comes with the features Henning was talking about. The short answer is no, but it's something we want to. HENNING SCHULZRINNE: This is an open invitation. We may not be able to do everything, but if you have things that you worry about, if you think something is not being deployed as widely or is being suppressed by intentional or unintentional actions, that's useful for us to know. I would be happy to talk to you about that. We can talk offline. I do believe there was at least one IMC paper that looked at ICMP options a few years ago. We have very little visibility into what survives the trip across the internet these days. 6. IAB Open Mic Bernard Aboba introduced the IAB: - Bernard Aboba, IAB Chair - Jari Arkko - Marc Blanchet - Mary Barnes, IAB Executive Director - Ross Callon - Alissa Cooper - Spencer Dawkins - Joel Halpern - Russ Housley, IETF Chair - David Kessens - Danny McPherson - Jon Peterson - Dave Thaler - Hannes Tschofenig Bernard Aboba noted that this plenary was a big change from the past, since the administrative portions had been condensed and combined with the technical plenary. He called for a show of hands for whether we should do this again. Many hands were raised in favor, with a few hands raised in favor of the old style. Lars Eggert (as IRTF Chair) and Heather Flanagan (as RSE) joined the IAB on stage for the open microphone session. A summary of the open microphone session follows: BERNARD ABOBA: We are now opening the mic to questions for the IAB. LEE HOWARD: I noticed--and this isn't just the IAB report--but all the presenters in the Administrative Plenary side said they would keep it short because it is their presentation is boring. If you think it's boring, don't waste our time. So why do it at all? DAVID KESSENS: Some presentations have to be done that are boring. Think about the IAOC presentations. This is the money that's being spent on the IETF's behalf, and we should be interested in it, despite that the presentation itself is boring. We can't help it. But some of the financial stuff you just need to know about. There's certain stuff that has to be done and trying to put it in a little corner somewhere doesn't work. We need to be open and transparent; that's how this community works. ERIC RESCORLA: Obviously, this is information that has to be public. The question is why it has to be presented verbally. BERNARD ABOBA: Is there any particular thing that you'd like to hear less of? ERIC RESCORLA: I'd like to hear less of all of the reporting. I'd like to see a URL for the reporting, and open mic other than that. JON PETERSON: There could be more hums to see if the community is interested in that. BERNARD ABOBA: That's not a bad idea. But I'd hate to be at the IAB plenary and make judgements on the IAOC. So just in terms of the IAB slides--we had 3 slides this time. How many of you would like to see 0? (Noone raises their hands). We'll take that as confirmation. SPENCER DAWKINS: Back to Lee's point, I feel your pain--but you don't actually want the IAOC report to be exciting. [Laughter from the audience.] JOHN LEVINE: I'd like the report to be either longer or shorter. If it's just going to be, "Here's what's on the web page," it could be shorter. On the other hand for the IAOC, I was intrigued that the first page said our revenue was up and costs were down, and we had this $500,000 surplus. And the next slides said, "We're probably not going to raise the price." Some analysis would have been helpful here. RUSS HOUSLEY: Can we talk about this with the IAOC on the stage? JOHN LEVINE: Sure. MARCELO BAGNULO: Maybe it would be better to do technical plenary first, and admin afterwards so that the people who are not interested can leave and look at it offline. JOEL JAEGGLI: One thing we forget is that the consequence of this organizational function is that instead of being run by Bob Kahn or CNRI is that it's actually being run by the people right here in the audience, and the IAOC, and the people that we appoint. So, the fact of the matter is that it terrifies me that people like Eric Rescorla are responsible for running this organization--but that said, he's not doing that bad of a job. [Laughter and applause.] JOEL JAEGGLI: We have the cookies, we have an administrative plenary which is tolerable, and we have a process that hasn't caused the IETF to go bankrupt. It needs a certain minimum amount of engagement from the people in this room, because some of you will be serving on the IAOC in the future, some of you will be serving on the IESG, and some of you will be serving on the IAB. You need to be engaged here, even when it isn't interesting. [Applause.] DANNY MCPHERSON: Back to Lee: most of what we do isn't sexy. This is the "exciting" boring stuff. [Laughter.] LEE HOWARD: Joel, I see no need to make threats that some of us will be serving on these various bodies in the future. My comment may have come off as flip, and maybe was misconstrued. My point was really about the apologies that the presenters were making. There's information here that is interesting, and maybe we should focus on the information and not the data. It is interesting how this stuff happens. Making sausage is neat stuff--if it's your thing. I'm not saying we shouldn't be doing the things that need to be done, but that this can be interesting, and can be made interesting by saying, "Look, here's the change in fees based on the analysis that we're talking about." This is good stuff. THOMAS NARTEN: What would you rather have: boring presentations that are short during the plenary, or really exciting stuff on the website that no one reads? ERIC RESCORLA: Hypothetical question: you guys all sat through that. Could you pass a test on this stuff? Could you reproduce any figure other than the $500,000 one? Because if you can't, this was a total waste of time. This would be fine if people were assimilating the information in any meaningful way. But they're not; they're just sitting through it, zoning out. Ask yourself if you could reproduce any significant piece of information you've just heard. And if you can't, then it was a waste of your time. LUCY LYNCH: Data point: we held an interim meeting. We had 36 attendees, plus 23 remote [attendees]. Work got done. It cost us $18,000. It wasn't all that successful. Data points from the slides. [Applause.] KEITH DRAGE: Just a request: for the reports that you don't give, like the RFC Editor--can you put them in the meeting materials rather than just links in the slides? JOHN KLENSIN: I didn't sit through this and zone out. I found it a very interesting and useful presentation. I have certainly sat in other plenaries and zoned out. Part of your obligation as the IAB is to present these things in ways that are useful to us so that we know them even if we don't want to know them. We need to know them anyway. If some of us are so omniscient that we don't need any of this information, then we shouldn't be spending more time in the plenary; we should be spending more time in the bar. But the rest of us actually do appreciate at least some of these presentations. They may be important to some of us who don't appreciate them but may need them anyway. I personally believe that we have gone downhill since we were using more plenaries which contained more information which was useful to large fractions of the community even if we didn't think we needed to know it. PHIL HALLAM-BAKER: I'm getting wrapped into this meta-meta discussion. We're currently having this Internet Governance Forum going on. All of the pre-meetings going on ahead of this ITU WCIT thing. There's some serious issues there that need to be discussed, and given the people here in this room, I would liked to have been discussing those. Regardless of what you may think that process has to do with this organization, another way that you can look at this process is that the plain old telephone system [POTS] is dying. And it's taking the institutions of that system with it. Those systems are linked into governments. My company has been attacked by a state actor. We've had another member of the CA family attacked by a state actor. We've had a very large corporation attacked by its own government--by the government of the country that we're now in. There's a lot of serious issues that I think we should be addressing. BERNARD ABOBA: We did have a proposal for a plenary which we didn't do this time, but which we could do in the future, about the issues of transitioning from POTS to IP. How many would like a plenary on the subject Phillip just mentioned, the death of plain old telephone service? [Lots of hands up in the audience.] BERNARD ABOBA: Wow, a significant audience. Okay, thank you. LESLIE DAIGLE: I would like to say that I did get information out of the brief administrative reports presented earlier, and I found them useful. The part I wish had been shorter was the rehashing of the plenary length; I don't think we've learned anything tonight about what the plenary length should be. The substantive point I want to bring up is in terms of the NomCom Chair's report, the articulation of what the NomCom is looking for in terms of useful input was very much focused on the IESG role. Everyone here who is doing any work is interested in making sure we have a well-constituted IESG. And in most cases, people have had enough interaction with the IESG that they have an idea of what does and doesn't work in the IESG. In a lot of cases, this [plenary] is the only interaction the community has with the IAB and IAOC, and I don't think that the community as a whole has a sense of what would or would not be useful in terms of skills and people represented on the IAB and IAOC. So I would like to hear from the NomCom Chair if he's still around. MATT LEPINSKI: If I had a brief articulation of what feedback would be useful for IAB and IAOC positions, I would have given it. I think there is a lack of clarity in the community as to what makes good IAB and IAOC members, and what the NomCom should be thinking about in regards to IAOC and IAB members. If you have any information about what you think the NomCom should be thinking about--we're talking to as many people as we can this week, and we'd love to hear from you. But I don't have a clear, concise, "This is what we need to know" statement. That's a problem in the community, and I don't know an easy way to solve it. LESLIE DAIGLE: It's not a new problem, or a problem specific to this NomCom. I'm so glad we're having this conversation, but could we have the discussion earlier next year? Because it would be really good to get some of that clarity and get people focused on what we need here. HANNES TSCHOFENIG: The charter of the IAB is laid out in RFC 2850, and it lists out the functions. Everyone has a different perception about the importance of the different functions. Some of them are about liaison relationships and interacting with other organizations. Others are more on the architectural oversight function, which is topics that concern multiple different working groups or even different Areas, or concern work that isn't even happening today in the IETF that may be useful to be done in the IETF. The plenary we heard today was one such topic, where we reached out to people outside the community to see if there is work to be done, bringing new blood to the IETF. Different people have different perceptions on what they would like their efforts being focused on. BARRY LEIBA: Following up on that last bit: the IAOC in particular requires the sort of background that we don't usually see people in here. Is it at all useful to hear how a person behaves in a WG when you're looking for contracts and business backgrounds and that sort of thing? MATT LEPINSKI: Yes, it's difficult for the NomCom to get input on the IAOC because of the way that most of us interact with the people who end up getting put on the IAOC. The NomCom is happy to receive an input, but with regards to the people that we're going to put on the IAOC to oversee the finances of the IETF and its contracts, it is not so useful to know how they performed in a working group. JOE HILDEBRAND: One thing that makes the NomCom's job difficult is that we don't periodically have a conversation about why we're here. What are we doing, what do we mean? Why do we have an IETF, why do we keep coming? Is it for the cool trips? Is there something we're trying to accomplish as a whole? And because we don't have that conversation, and because we have a strange process (as compared to other organizations) for the choosing of our leaders, I think we should periodically have that discussion about why we're here. If we knew that with some specificity, it would be easier to pick the people who are trying to forward that goal. OLAF KOLKMAN: If the NomCom and the community don't know what the IAB and the IAOC are doing, then maybe the plenary presentations are too short. [Laughter.] OLAF KOLKMAN: Or maybe we have not been communicating well enough what we do. I don't have good ideas about how to improve that communication, and I don't think the big plenary hall is the place to have that conversation. If you don't know what the IAB and IAOC are doing, please ask. It's part of this community, and this communication needs to be clear. Set your requirements. BERNARD ABOBA: I'd like to ask an additional question of the audience. We have had suggestions from some members of the community that maybe we could use the remote presentation facilities from time to time. If we were to have a technical chat, some introduction to the IAB, how many people would actually log into that and participate? [A few people in the audience raise their hands.] BERNARD ABOBA: Well, maybe our remote participation system could handle that, if we're only talking about 12 or so people participating. Not a very big load. [Laughter.] PETER SAINT-ANDRE: I don't know if Eric has been paying attention in working group meetings, but the plenary has no monopoly on folks zoning out. [Laughter and applause.] PETER SAINT-ANDRE: We need to focus more on the issues that require real-time interaction. We should use this high-bandwidth connection to discuss what's really important. [Applause.] LESLIE DAIGLE: I wanted to follow up from Hannes's characterization of what the IAB needs. I'll start by responding to Bernard's suggestion about open tech chats with the IAB. This is going to come across as sounding a little rough, so please don't take it the wrong way, but I don't think anyone here wants to get to know the IAB personally. I mean, I am fortunate enough to know many of you personally, so I have that advantage. My point is more about what can you do? What are your skills that you need to get your work done? That's what the NomCom needs to know. The IESG, when you fill an IESG seat, you don't just find who is the brightest spark or most intelligent person. You're looking for someone who can sort out actual working groups. And we're all familiar with that process because we've been on the other end of it. We're not on the other end of the things that the IAB does. While I appreciate that there are lot of policy things coming up, I think that it's not a question of finding particular expertise and making this the panel the panel of Internet gods. I think this group needs to be constituted in a way that you all work together, and that you can understand something outside of your own particular, which means that you're technical generalists at a certain level. That you care about it, and you care about finding the right answers--that's not something that's going to convey in a tech chat. BERNARD ABOBA: I would point out that the IAB has provided a very detailed job description to the NomCom, and for the first time in a long time, the NomCom has not asked questions about it. HANNES TSCHOFENIG: Leslie, you forgot one really important part, which is that the person has the time to do the work. Because that is generally the show-stopper, both on the IAB and the IESG. If you don't have the time, then it really doesn't work. DAVE THALER: I think as Bernard was saying, the job description contains some things, and I don't know how many people who provide feedback about IAB members have actually read the job description, but that's at least some answer. It's a variety of skills that no single person would have, but collectively across the IAB, different skills that are necessary to have within the IAB: the ability to look at a problem you're not an expert on and make an insightful comment. Or things like being aware of the politics of different situations, like being aware of how different SDOs work, or how different governments work, or the effects of governmental-run organizations. Just being aware of the politics is another whole aspect, and being a generalist there is also very useful. Collectively, these are they types of skills that are useful. BERNARD ABOBA: And the job description has been changed substantially in the last three years, in particular to emphasize more leadership skills. So thank you all very much for your feedback. 7. IAOC Open Mic Bob Hinden introduced the IAOC: - Bernard Aboba, IAB Chair - Scott Bradner - Dave Crocker - Marshall Eubanks (not present) - Bob Hinden, IAOC Chair - Russ Housley, IETF Chair - Ole Jacobsen - Ray Pelletier - Lynn St.Amour, the ISOC President/CEO (not present) A summary of the open microphone session follows: MARC BLANCHET: We used to have live transcription during the plenaries, and that's disappeared. Is there a reason why? RAY PELLETIER: Originally we had the transcriber here because they were assisting someone here with a disability who could use that service. And then we broadened that to include the occasional other meeting, and then the plenaries. And while that had some popularity, it's been gone for a couple of meetings now, and nobody complained. It seemed like perhaps it wasn't of value to most folks. JOHN LEVINE: You have a substantial surplus this year. So why are you asking ISOC for less money rather than reducing meeting fees? ISOC appears to have no shortage of funds, and for those of us who pay our own way here, while the fees are not unreasonable, they are not low. What's the overall logic for some of the meeting fees? BOB HINDEN: I'm not sure exactly how we got to the $650 number. It was increased about 3 years ago from $635, so we've tried to keep it fairly consistent. But actually we're asking for more from ISOC for next year than we did for this year. We are actually asking for more money, not less. SCOTT BRADNER: The ISOC board took an action to increase funding in order to avoid raising meeting fees. RUSS HOUSLEY: ISOC helps us find sponsors, and we had a lot of sponsors this year, so that's where the additional revenue is coming from. JOEL JAEGGLI: I'm on the recall petition, but one question I have after reading BCP 101 and RFC 3777, is that when you're replacing mid-term people once you've created a vacancy, which the recall process does, the IAB reviews IESG appointments, and the ISOC board reviews IAB appointments, but no one reviews the IAOC appointments? RUSS HOUSLEY: The IESG is the confirming body for NomCom-selected IAOC appointments. SCOTT BRADNER: See RFC 4071, section 4. JOHN KLENSIN: We make a big fuss about cross-Area review and cross-Area understanding. Any time a mic line has to be closed, it's an indication the plenary is too short. I'm old enough to miss the Monday morning plenary which would highlight the important things that were going on in the week. I think we are damaging ourselves as a community if we keep shortening the portions that allow input from the community, discussion from the community, and technical review of important work across Areas. BERNARD ABOBA: It's an interesting observation. To some extent, I was sitting up here telling people about things that had already happened. That did feel a little awkward, as opposed to Monday night when I could tell people about things that were going to happen. JOHN KLENSIN: I think that the move of the IAB plenary to Monday and the ability to use that slot for forward-looking stuff was a real improvement. But again, I'm one of those fuddy-duddies who thought the short Monday morning plenary was a good idea. It's not a matter or reporting on what happened during the week; it's an opportunity to let all of know, if we have spare cycles, where we ought to spend them. And if that bores some people, shame on them, because what's important here is our ability to work as a community across Areas and across levels of the stack. DAVE CROCKER: I am confused about the logistics of what you're suggesting given the number of Working Groups we have. I'm unclear about how much time you want and what you want done during that session. JOHN KLENSIN: I think that kind of review is more important than another Working Group. I think if we cannot handle a Working Group in terms not only of managing and having people come to the sessions, but of understanding as a broad community what that Working Group is doing, then we have too many Working Groups. So my tradeoff isn't more Working Group time. It's making more strategic decisions about what we need to do. DAVE CROCKER: But what are you looking for on Monday morning? JOHN KLENSIN: I think the Steering Group is supposed to act like a Steering Group. If the Steering Group is able to say, "Well, this Working Group has an impact across Areas on this thing on the internet, and maybe it's time we get a presentation from them," that's useful. DAVE CROCKER: A selective hot seat for salient working groups? JOHN KLENSIN: If that's necessary. For some we've been using these cross-Area reviews at Last Call as a means to get input from those which would not normally occur. That's too late. MATT LEPINSKI: It would be kind of cool on Monday morning to have the IESG members give 5-10 minutes each on what they see as the important things that will happen during the week. Or just a couple of sentences from each IESG member. That would help give everybody a heads-up about things that are happening in other tracks. RUSS HOUSLEY: So, on Sundays the IESG and IAB talk about hot spots. Maybe we could do an experiment in Orlando and hold that discussion in plenary. JOHN KLENSIN: I think that would be useful, but it would have to be no later than Monday night. RUSS HOUSLEY: Right. SAM WEILER: I am not happy about response to John Levine's question about meeting fees. If a more substantive answer is not more immediately available, I would like you to take an action to review those fees, and if you're not lowering them, respond back to the community with an explanation. RAY PELLETIER: Just for background, the fees $635 for 2008, 2009 and 2010. And they are $650 for 2011, 2012, 2013. SAM WEILER: That doesn't answer the question. RAY PELLETIER: Just a point of information. ERIC RESCORLA It's not fun to admit you're wrong, but I hear it's good for you. I just found the discussion of the meeting fees to be quite illuminating, and we wouldn't have had that if we hadn't seen the financial presentations, so bring it on. [Laughter and applause.] ERIC RESCORLA: I just got back from W3C where the last-minute meeting fee is 135 Euros a day, so I'm feeling okay. BOB HINDEN: Thanks for the feedback. DANNY MCPHERSON: Just another data point: I know we [Verisign] have sponsored stuff at the last couple of IETFs with the resources we had available, and that some of those sponsorships only came in a few weeks before the meeting, so the IAOC can't count on that money coming in. They have to budget to manage this long-term. TIM SHEPHERD: I think something unfortunate happened with the meeting fees discussion, and I think there's a lot of misunderstanding. My theory is that I think some people might think that the IETF is running a surplus. But from what I understand from sitting through similar financial discussions at previous meetings is that the IETF is actually running a substantial deficit every year, and that deficit is being made up by ISOC underwriting the activities of the IETF. SEVERAL IAOC MEMBERS: That's completely correct. DAVE CROCKER: It depends on whether you think the ISOC revenue is a natural part of the IETF revenue. PAT THALER: Having been treasurer of another large standards organization, one thing I'll note is that this year we had 3 North American meetings, and that's not the desired case or the usual case. And frankly, the costs in North America, because of the way the hotels are set up to charge for their meeting space, tend to run lower. It's hard to extrapolate from this year's experience what next year will be, and I think that is important to consider. BOB HINDEN: That is correct. You'll see that in the back up online, if you look at the years where we had more meetings outside North America, the costs do go up. And to the previous statement, we run approximately a $2 million deficit every year. It's being made up by ISOC, but it doesn't show up on the revenue line. 8. IESG Open Mic Russ Housley introduced the IESG: - Ron Bonica, Operations and Management Area - Stewart Bryant, Routing Area - Gonzalo Camarillo, Real-time Applications and Infrastructure Area - Benoit Claise, Operations and Management Area - Ralph Droms, Internet Area - Wesley Eddy, Transport Area - Adrian Farrel, Routing Area - Stephen Farrell, Security Area - Russ Housley, IETF Chair, General Area - Brian Haberman, Internet Area - Barry Leiba, Applications Area - Pete Resnick, Applications Area - Robert Sparks, Real-time Applications and Infrastructure Area - Martin Stiemerling, Transport Area - Sean Turner, Security Area A summary of the open microphone session follows: LUCY LYNCH: There isn't another natural place in the plenary to say this. I've seen a number of IETF 55 shirts this week. On the bottom of that T-shirt, it says "IPv6." Itojun did that. Itojun was a volunteer on the NOC team at 55. It's hard to believe its 30 meetings on, and I still miss him. So thank you, Itojun. [Applause.] ANDREW SULLIVAN: There was a recent incident where there was a suggestion for somebody to make a judgement call. And the response on the mailing list was, "Oh, we better not do that; we should exercise one of these rules instead." We have a current discussion going on about the Note Well, and it's turned into a 300 message cluster of people micromanaging the discussion of exactly what we can put up in slides. I am extremely concerned that we need the ability for someone to make a judgement, and make it stick. I'm wondering if you have any suggestions on how to make that work in the future? RUSS HOUSLEY: Your question is how to stop bikeshedding? ANDREW SULLIVAN: Well, yes. PETE RESNICK: One of the things that generates the bikeshedding about process is driven by an idea from the IETF plenary (the main body of everyone) that we are authorities that need to take charge. The decision has to be immutable, and if we're going to be put in that position, then everybody wants to make sure that the rules for us doing that are really tightly defined. On the flip side, because of that feeling, we are loath to engage in the discussion on the list, for fear of overly influencing the direction of the discussion. There's a power imbalance now, and a pretty serious one, that when I started with the IETF wasn't here. Part of this is a cultural problem, that we're being asked to be in a very legalistic, authoritative place, and the dynamic has gotten strange. Some of it is, you guys shouldn't worry so much about the rules, but also, push back on the judgements. Does that make sense? ANDREW SULLIVAN: Yes, it makes sense, and that's exactly why I want judgement rather than more rules. It seems like every problem that we have, we immediately go into protocol-building mode. We're always fighting the last war. I am terrified that eventually what we're going to have is all process, and do no actual work anymore. RUSS HOUSLEY: This is exactly why open mic time is important, because this is where the community holds us accountable. HARALD ALVESTRAND: I am just slightly sentimental. I was looking at the slides I did at the plenary in Atlanta 10 years ago. One slide says, "The IETF is a success in producing high-quality, relevant standards for the internet and allowing open participation and fair sharing of ideas. The leadership is providing a unified vision of the internet standardization effort, and in using the internet to create the internet." The next slide says, "The IETF is a failure. Work is slow, output mediocre, and irrelevant to the real problems facing the internet. Decisions are taken by back-room deals, intimidation and mob psychology. The IESG imposes random mandates that are irrelevant to the problems at hand. People are leaving in disgust in droves." Some of you were at that plenary. I am quite happy to see that compared to that plenary, plenaries are reasonably tame now. [Laughter.] HARALD ALVESTRAND: Thank you for making the stuff that needs to be done boring to the rest of us. We need to get our work done and not have that kind of dramatics. RUSS HOUSLEY: Thank you. DAVE CROCKER: Back to Andrew's point. I think there's a couple of things make a huge effect on the bikeshedding and devolution into minutiae. Yes, we all have the tendency. Welcome to the world of engineering. We know how to manage processes in Working Groups, and we don't do that management in the IETF list. We don't have someone who tries to move the conversations towards convergence. We tend to let tiny numbers of people effectively have a veto, and rough consensus doesn't assign that kind of power to individuals. For topics that don't formally require rough consensus, let me suggest that there is a difference between discussion (which is good) and sense of obligation to have a clear rough consensus (which is bad for those topic where it's not required). The thing that we probably do need when we don't need rough consensus, is reasonable support. If you don't have a fair amount of community support, it's probably a bad decision--but that's different from saying we require rough consensus. This distinction between when we need rough consensus and when we don't might help a lot. PETER SAINT-ANDRE: Having sat up there myself, I'd like to point out that those people up there [on stage] aren't our leaders; they are people serving in temporary leadership roles. They are people just like us. JOHN KLENSIN: I consider that we don't have a plenary like the last one in Atlanta as a bad sign of community disengagement. I'd like to see more controversy around here, because I think the controversy exists in the community. And there's a sense of acceptance and reassignment that one can't do anything, that's a bad sign. We don't have enough recalls, we don't have enough controversy, we don't have enough appeals. That's not to say people are behaving badly, but that the community doesn't feel it has the leverage to push back because pushing back gets bad results. JOE HILDEBRAND: A disengaged community produces bad work, and I think we've been guilty of that, and I think that's why it's hard to get people involved in new work. For instance, we have people participating on mailing list who live here in Atlanta who we couldn't get to come to the meeting physically because they weren't interested in our level of back-biting and production of low-quality output. And I think that's a shame. We need to think about who we are, and what we're doing, and why we're doing it. ADRIAN FARREL: Isn't it possible that the people on the list chewing the ideas are the ones who cause the disengagement by the others? It's the engagement that causes the disengagement. JOE HILDEBRAND: That may be. And part of it is that we don't teach other to argue well. We end up arguing about the wrong things, and we do it in a way that discourages people from having opinions. Winning isn't important; what's important is building a better internet. PHIL HALLAM-BAKER: I find it strange that we're discussing how disengaged people are these meetings. On the one hand, we've got these people who are not engaged. On the other, we've got these people who are saying that they wouldn't want to be engaged. I think we've got to decide whether it is more important to get people engaged who are interested in making the internet better, or whether it is more important to preserve the institutions in the form that they have evolved to over the last 25 years. PETE RESNICK: I don't know that there's anything contradictory in those views. What I think I heard Joe and other folks talk about in terms of engagement is that there is the way we deal with each other internally, and then there is how the rest of the world deals with us. They are related, no doubt, but I do think that what folks are talking about is more about how do we make that communication work internally. PHIL HALLAM-BAKER: I think the two are linked, and it points to a systemic issue. HARALD ALVESTRAND: Ten years ago I expected about five different revolutions to have to happen in order for the IETF to survive the next 10 years. I think we've had about two of those. I am concerned that we don't have a process that scales. If someone asked us to come up with good reason why we had to produce 30% more output in a year because it was needful, I don't think we could do it. I think we have serious problem in scaling and in quality of reviews. That we've kept the machines churning and produced output of such quality is a credit to the leadership; we just haven't had the revolutions yet. RUSS HOUSLEY: I think that's supported by the time commitment by the leadership positions. KATHLEEN MORIARTY: While I don't disagree with the other comments that have been made, I want to say that from my experience, the level of detail and transparency in the IETF is tremendously helpful, especially as I've brought new people into this forum. I greatly appreciate that. LEE HOWARD: Think it's great we can have such an active discussion about how disengaged we are. [Laughter.] LEE HOWARD: I don't hear grousing about what ADs are doing. I think the reason there aren't so many fireworks is that the IESG is doing a great job. RUSS HOUSLEY: Thank you. MURRAY KUCHERAWY: In comparison to some of the other people who have been at the mics, I'm a relative whippersnapper. However, I'd say that in the time I've been here, there have been big improvements in all the things we've talked about, the accessibility of the IESG. I'm fairly satisfied in the way that's going. There's another organization I attend regularly which hasn't documented any of its process, and I frankly use the IETF as a model for it sometimes. They're about one- fifth the size of this organization, but I'm much more comfortable in this environment than it that one. PETE RESNICK: I think Murray is right about the engagement bit. That is, once you get to know us, it's easier to engage us. But most of the community is not engaging us at that level. One of the reasons people become chair is because they engage us, and vice versa. The people who come up to the mics are the ones who are engaged. There are a lot of people roaming the halls who have not gotten up to the mics. RUSS HOUSLEY: Six years ago the IESG had a breakfast meeting on all six mornings of the IETF week, and now we have only three, which allows the IESG to be out there having breakfast with the community and have those two-minute conversations that can help avoid a train wreck. PETE RESNICK: Although I will admit to spending half those breakfasts on topics in a specific Working Group and not the general engagement. So, nose whacked with newspaper, appreciated. PETER SAINT-ANDRE: I'd like to ask Harald the question about what those other revolutions are? HARALD ALVESTRAND: One of the suggestions I made at one point was that the complete IESG never see a document unless there is a crisis. The review process doesn't scale. It's much more important for the people at the top of the pyramid to provide actual leadership rather than reviewing documents. Reviewing documents is of course critical, but these are not the people to do it. RUSS HOUSLEY: I distinctly remember the slide you put up when you stepped down as chair, and it was not a pyramid--at least not a right- side-up one. HARALD ALVESTRAND: The way Brian Carpenter described the way you know you're IETF Chair is because everyone sits on you. One of the other revolutions I suggested was cutting the IESG in half, because the group is too big to communicate efficiently with itself. That didn't go anywhere; we've since added a new Area, which meant adding two ADs. But these are mechanisms, not revolutions. I would like to have the IETF be an organization that eats other standards organizations. When people see that running an organization is no fun, they should come to us and ask us to take over. And the IETF should be in a position to say yes to that. We've kept the machine ticking, and that's important, but we can't continue this way indefinitely. JOE HILDEBRAND: I just wanted to get one positive note in. I do think there are a lot of really strong individuals, and a lot of people that I've learned a ton from over the years. One thing for the people who are in leadership or who have been around for a long time and have some of the institutional knowledge, they should really be asking themselves if they have enough people that they're mentoring. If we want to culture to maintain itself over the years, please keep taking new people under your wings. JARI ARKKO: As we do more and more engineering, the systematic function gets harder and harder. At the same time, some other organizations are pushing us on speed (like WHATWG). We do have to figure out how to get back to more "running code" principles. The danger is that certain efforts will go other places if we're not fast enough. RALPH DROMS: If we're going to speed things up and look more at running code, then we need to be careful not to lose our breadth. [ENDS]